Vertical Relationships and Competition in Retail Gasoline Markets: An Empirical Evidence from Contract Changes in Southern California
Social Science Research Network
In a paper in the March 2004 issue of the American Economic Review, Justine S. Hastings studies the acquisition of a sizable independent gasoline retailer, Thrifty Oil Company (Thrifty), by a vertically integrated refiner/retailer, ARCO. She employs a difference-in-differences approach on a panel of station-specific prices to examine the price effects of this transaction at nearby competing stations. She finds that the pure rebranding effect of the transaction increased retail gasoline prices
... five cents per gallon at competing stations. This effect is identified separately from changes in horizontal concentration and differences in the degree of vertical control, for which no effect on prices was found. The size of the estimated price effect-five cents per gallon amounts to a 50 percent increase in retail margins-along with a desire to better understand the mechanism behind it, motivated us to revisit Hastings's analysis. 1 Being unable to acquire her data, we used an alternative source, which in aggregate is very similar to the original dataset. While there are differences between the two datasets, the five-cent effect is large enough that we would expect to find an effect of a similar order of magnitude. Ultimately, however, we find an effect of tenths of a cent per gallon, which is of little economic significance. This finding is robust to using various subsamples and analysis of higher-frequency data, unavailable in the Hastings data. 2 Our empirical results cast doubt on whether ARCO's acquisition of Thrifty led to higher prices. 3 1 The average price of regular grade gasoline, without tax, to end users in California in 1997 was 86 cents per gallon. The average wholesale price of regular gasoline in California in 1977 was 76 cents per gallon (Department of Energy, Energy Information Service, gasoline prices by formulation, grade, and sales type; available at http://tonto.eia.doe.gov/ dnav/pet/pet_pri_allmg_c_SCA_EPMR_cpgal_a.htm). Hastings's estimated price effect measured as a percentage of the combined refining, wholesale, and retail margin of gasoline is sizable as well, since the average price of crude in California in 1997 was 38 cents per gallon (Department of Energy, Energy Information Service, California Crude Oil First Purchase Price; available at http://tonto.eia.doe.gov/dnav/pet/hist/f005006__3a.htm). 2 In addition to the specifications described herein, see Christopher T. Taylor, Nicholas M. Kreisle, and Paul R. Zimmerman (2007). 3 Our results also cast doubt on the underlying model of consumer preferences (differentiated products with consumer brand loyalty) for which Hastings finds support in her data. Even if this model accurately depicts consumer behavior, we note that its welfare effects are ambiguous because the introduction of a new brand increases gross consumer utility. Hastings describes how rebranding can soften price competition but makes no claims as to welfare effects.